We're evaluating whether Workflow-Core is suitable option for us as a workflow engine to drive an end user automation feature to allow our users to create their own simple automation steps (eg, when x happens, then do y then z). Our approach so far is that when a user sets up a workflow, we create and register a workflow definition to represent it and it seems to be working. But with this model, there will likely be many thousand Workflow Definitions as users set up their own automation rules.
One thing we've noticed is in the WorkflowRegistry implementation (below) it loads all definitions on startup and stores them in a concurrent dictionary in memory. We're wondering whether there may be a ceiling we hit here if we have a lot of definitions.
Hi team
We're evaluating whether Workflow-Core is suitable option for us as a workflow engine to drive an end user automation feature to allow our users to create their own simple automation steps (eg, when x happens, then do y then z). Our approach so far is that when a user sets up a workflow, we create and register a workflow definition to represent it and it seems to be working. But with this model, there will likely be many thousand Workflow Definitions as users set up their own automation rules.
One thing we've noticed is in the WorkflowRegistry implementation (below) it loads all definitions on startup and stores them in a concurrent dictionary in memory. We're wondering whether there may be a ceiling we hit here if we have a lot of definitions.
https://github.com/danielgerlag/workflow-core/blob/ab97ce9c24670b7f9810e8d0da94141fe771c41a/src/WorkflowCore/Services/WorkflowRegistry.cs#L11-L35
So I guess my questions are:
Thanks, and we really appreciate your time looking over this.